Michigan Tech Research Institute (MTRI)

Remote Sensing-Based Water Color

Michigan Tech Research Institute (MTRI) and the Nansen International Environmental and Remote Sensing Center (NIERSC) of St. Petersburg, Russia have developed a new algorithm for the retrieval of color producing agents (CPAs) for the Great Lakes. The algorithm is applicable to all coastal and inland waters, and operates on either multi-spectral (MODIS, SeaWiFS) or hyper-spectral (Hyperion, AVRIS) data and produces estimates of chlorophyll, dissolved organic carbon, and suspended minerals. The algorithm has undergone a preliminary validation using both dedicated and historical /in situ/ water chemistry measurements. See Pozdnyakov, Shuchman, Korosov, and Hatt (2005) for a more extended description.

Algorithm chart.
A flow diagram for an advanced operational algorithm for the retrieval of CPA concentrations in Case II waters. Satellite imagery is pre-processed, then the subsurface remote sensing reflectance spectrum is calculated for each pixel. These spectra serve as the inputs to an array of broad-band neural networks, which effectively narrows the range of initial concentrations for the Levenberg–Marquardt multivariate optimization procedure, which solves the inverse problem to retrieve the final CPA concentration vector.

The new, fast-operating algorithm is based both on a previously developed hydro-optical model developed for Lake Ontario (Bukata, et al., 1995) and a combination of the Levenberg-Marquardt (L-M) multivariate optimization approach with a neural network (NN) emulation technique. The NN emulator provides the L-M procedure with reasonable initial constraints, which makes the final solution of the L-M procedure more accurate. It also speeds up the retrieval algorithm performance significantly.

A seven-year time-series analysis of Lake Michigan was reported in Shuchman et. al. (2006). It used the results of the algorithm to identify seasonal biogeochemical events, and well as to predict the results of climate change in the Great Lakes basin. The paper won the Chandler-Misener award for 2006, given by the International Association for Great Lakes Research (IAGLR) for outstanding research about the Laurentian Great Lakes.

Three maps with colored areas denoting water quality.
Example outputs of the CPA algorithm for an image of Lake Michigan: chlorophyll concentration (left), colored dissolved organic material (center) and suspended minerals (right).